F23: Just Got A New Camera - Sony F23

Thanks for chiming in Mitch. I'm really excited to be playing with the very best digital cinema equipment money can buy - even if I'm doing it 8 years late. I know in this forum there's been some non-sense discussion of blowing the F35 up to 4k, but I've done some simple comparison between the FS7 at UHD and the F23 blown up to the same - and the FS7 confers virtually no advantage in true resolution on that test.

The tests I'm doing on color response from the F23 paired with DigiPrimes... without a doubt, the best combination I've ever had my hands on.

And I'm pleased to be pairing it with your company's O7Q+.

Mitch - I'm having some trouble googling original cost on DigiPrimes. Was it like $30k per each lens? And what about the F23? Sans SRW-1? $400k with SRW1?

Thanks!

Also... I'm enjoying 4:4:4 DPX, but what's the timeline on 4:4:4 ProRes?
I think the DigiPrimes were a little less than that, but not drastically so. At least $20K each.

I'm fuzzy on the F23 price, but I'm thinking it was something like $180-$240K. But what's a couple dozen grand between friends?

As for compressed 444, the timeline is "as soon as we can". Some things are at least partly out of our control. :-)
 
Very busy week for me. Will be doing two productions with the F23s this week while the F35 sits out.

Yes - the "mystery stills" were the F23.

Dropbox gave me some issues, but the files may be available again. I'll try and make more clips available sometime soon. Little bit tied up right now.

Photoshop - all I did was a simple 200% enlargement. Photoshop has choices for how you resize, with different algorithms for scaling down or scaling up. Essentially, when choosing an enlargement it tries to make sensible interpolation choices. Keeping the enlargement math simple (200% - 1 pixel becomes 4 pixels, exactly) produces superior results to odd numbers that force less uniform scaling.

I have a feeling the the F23 does better with this enlargement that the F23 on the basis of the color fidelity afforded by the prism sensor. The RGB strip sensor isn't getting the same accuracy on color.
 
But, here's the thing - once upon a time, "4k" meant a film scan done at 4096 lines red, 4096 lines blue, and 4096 lines green - essentially, RGB 4:4:4 at 4k. But Red came in and touted a single chip bayer sensor at 4096 lines as "4k" and the definition got altered. So, the very best in 1080p that the F23 and F35 produce can be blown up to 3840x2160 in Photoshop and can then be difficult to delineate from bayer sensor UHD 4:2:2 acquired images.
To be fair, RED wasn't the first. The Dalsa Origin did this first and the Vision Reach Phantom was guilty of it as well, and both of those cameras were available years before the RED One.
 
To be fair, RED wasn't the first. The Dalsa Origin did this first and the Vision Reach Phantom was guilty of it as well, and both of those cameras were available years before the RED One.

Thanks. I was still in school those days, so I'm not always clear on those timelines.
 
The F23 looking comparable to the original RED one, or even things like the F5 really isn't such a leap.

Total MP count of the F23 is approx 6mp, total of the F5/R1 is around 8. This is where I get driven nuts by people wanting to shoot 4K ProRes - there was only 33% more data to begin with than good HD, but you're using 4x the space to store it (assuming equivalent codec).
 
The F23 looking comparable to the original RED one, or even things like the F5 really isn't such a leap.

Total MP count of the F23 is approx 6mp, total of the F5/R1 is around 8. This is where I get driven nuts by people wanting to shoot 4K ProRes - there was only 33% more data to begin with than good HD, but you're using 4x the space to store it (assuming equivalent codec).

More specifically, the F23 has 2.07mp for each channel - red, green, blue.
The 16:9 portion of 4k (being UHD) on a Red One, F5, FS7, etc is composed of approximately 2.075mp red, 2.075mp blue, and 4.15mp green. So, nearly identical in terms of color photosite except for the bayer sensor offering double green. Not to mention the better color sampling accuracy of the prism.
 
The F23 looking comparable to the original RED one, or even things like the F5 really isn't such a leap.

Total MP count of the F23 is approx 6mp, total of the F5/R1 is around 8. This is where I get driven nuts by people wanting to shoot 4K ProRes - there was only 33% more data to begin with than good HD, but you're using 4x the space to store it (assuming equivalent codec).

Are you getting the 6MP figure by multiplying 1920x1080 x 3 because it's a 3-chip camera? Because if you will do that, the EX1 was also a 6MP camera but yet nobody considers it one. It's a 2MP camera, just like the F23 has always been seen as.


It's different when all the pixels are in the same sensor as as with the RED or F5.


I understand the excitement of yet another previously out of reach camera now being affordable. But I believe, specially in this section of the forum, objectivity should take precedence.

As far as upscalling HD to 4K or UHD, I have done tests with the F35 using several methods. From Resolve to a Teranex unit and it looks superb indeed. Although I have not done a side by side I would also say it doesn't look much lower resolution if at all than these pseudo-4K cameras such as the F5 or F55 or anything outside of F65 and Dragon. Again without testing, I would risk going as far as saying it should match the in camera upscalling of the Alexa pretty well in definition. Not surprisingly though. Although the F35 supposedly records HD only the images are very well defined. It doesn't take two looks to see it is resolving far beyond 1080p. Actually, if you point it at a resolution chart you will confirm it is resolving between 2.4-2.5K. So from a definition standpoint it is not starting too far bellow the Alexa to upscale.


With the F23 I have never done any tests. But as far as I know the F23 was designed to be a true and whole HD camera. It's effective pixels are 1920x1080. So the F35 most likely resolves more. Again not surprisingly as it was developed to offer detail and quality closer to film, which is why it is so over-engineered.
But I'm pretty sure F23 upscalled would look very nice as well because the base image is so good. From a quality perspective, it would be hard to say something out of an a7s, GH4 or even FS7 and F5 in 4K would look better than anything from a F23 upscalled. It may even look sharper. But it will never match the color and image depth of the F23.


Naturally, the same is true for the F35.
 
Aaveum

Yes, I am doing the simple 1920 x 1080 x 3 math. I do not consider the EX1/3 to be in the same league because they are 1/2" sensors, completely different marketplace and tech.

The point is more that marketing was fiddling numbers from the beginning - ie. saying "oh look we have 8MP versus their 2MP" - that just flat out isn't that simple.

Until fairly recently nearly everything was mastered to HDCAM SR @ 1080P anyway (at least the stuff I was involved with), so it was a moot point - getting the best possible HD image was all that interested me and I felt the F35 did that.
 
Are you getting the 6MP figure by multiplying 1920x1080 x 3 because it's a 3-chip camera? Because if you will do that, the EX1 was also a 6MP camera but yet nobody considers it one. It's a 2MP camera, just like the F23 has always been seen as.


It's different when all the pixels are in the same sensor as as with the RED or F5.

Your comparison to the EX1 is deeply flawed. While the EX1 may have 6MP to work with for internal processing, it's best output is YUV 4:2:2. The F23 however, has perhaps the finest aligned CCD prism block ever made, and offers an RGB 4:4:4 output, meaning each of those individual 6.22MP photosites is readable in uncompressed 10 bit log from a 14bit ADC. Yes - the design of the prism is different than a bayer sensor - it offers higher color accuracy but lower luma resolution. But these are competent points. Your comparison might hold water if the images of the EX1 and the F23 looked anything alike - they don't.
 
So, super busy taking care of this weeks shoots - but having just done a concert style shoot with my F23s as Cam A and B...

Wow. Just wow. Wish I could share now, but will later. They handle stage lighting like a champ - even saturated stage led lights. In fact, the stage LED lights broke my A7S's SLog3 X-AVCs, but the F23's held it together. Also exceptional when thrown strobe like flash lighting events. Specular highlight handling was outstanding. I'm very pleased.
 
444 comes into its own under stage lighting too. Nothing like having the resolution of your scene halve (or quarter) when they go to all-red or blue lighting!
 
Guys,


Of course the EX1 is not in the sameleague as the F23 and I never said it was. What 2/3” HD camera is,besides it's twin the SRW-9000?


Viddovation, my comparison between theEX1 and F23 is not flawed. Because I never compared the two.


I was talking about “pixel countmethodology” if you will. It has nothing to do with the outputquality of a camera or the size of the sensor for that matter.


In the whole history of 3-chip camerasthe megapixel count was never really counted by multiplying the three chips. So the example (and not comparison) of the EX1 is indeed veryvalid here. But if you are not comfortable with that we can just usethe F900R instead and the same will hold. Nobody never considered theF900R a 6MP camera. It's a full HD camera and so a 2MP camera. Justas the F23 or SRW-9000. The quality of the sensor block, prism or whathave you doesn't change that.


If you want to create your ownmethodology to fit your plot, then it's something different. But it would be just anartificial plot device. :)
 
So, super busy taking care of this weeks shoots - but having just done a concert style shoot with my F23s as Cam A and B...

Wow. Just wow. Wish I could share now, but will later. They handle stage lighting like a champ - even saturated stage led lights. In fact, the stage LED lights broke my A7S's SLog3 X-AVCs, but the F23's held it together. Also exceptional when thrown strobe like flash lighting events. Specular highlight handling was outstanding. I'm very pleased.


This is the perfect job for a 3-chipCCD camera. Here they are unmatched. Perfect example of the oftenpreached but seldom followed “ The right tool for the right job”motto.


This is really an arena the F23 can besold well with just the right marketing approach. Pretty easy toshow a client a side by side between the F23 and a F5 or any otherCMOS camera, specially with a rolling shutter to clearly display theadvantages of your set up.
 
I think the point trying to be made is more this:

Is an F5 (for example), higher res than an F23? Yes, definitely.

Is it four times sharper? Absolutely not.

Bayer patterning does a much better job than it should, but it simply will not resolve the same amount of detail (if using a "4k" sensor for a 4k picture) as the triple CCD method. The only way is massive oversampling, eg. F35.

Note also Sony usually called the F35 a 6mp camera, possibly a typo.
 
Not sure which I'm more envious of.... the cameras or the glass. Wow.
Looking forward to seeing some grabs too. Just curious - have you used the DigiPrimes on your FS7 yet?
 
Not sure which I'm more envious of.... the cameras or the glass. Wow.
Looking forward to seeing some grabs too. Just curious - have you used the DigiPrimes on your FS7 yet?

The DigiPrimes won't natively fit the sensor on the FS7. You'd need a B4 optical enlarging adapter. I've not done that yet, and it's not a priority for me.
 
The DigiPrimes won't natively fit the sensor on the FS7. You'd need a B4 optical enlarging adapter. I've not done that yet, and it's not a priority for me.
That device would be the IBE HDx35 optical adapter. When it is used in conjunction with the DigiPrimes the results are gorgeous.
 
I shoot UFC Ultimate Insider on red one with HDx35 adapter. Works very well, except for 2.5 stop light loss. Production owns the camera and wanted to stick with red one, Jib shots need ENG super wide zoom lens so this is how we arrived at the combo. We are usually shooting at 1000ISO, 2.8 on lens and blasting the light on set. In terms of sharpness, it is just as good as when on a 2/3 camera.

Looks like another F23 is on eBay. For 4 years I looked for one, now in the last few months they are coming out of the wood work.
 
Last edited:
Back
Top